Photovoltaic array condition monitoring based on online regression of performance model

Photovoltaic (PV) system performance can be degraded by a series of factors affecting the PV generator, such as partial shadows, soiling, increased series resistance and shunting of the cells. This concern has led to greater interest in improving PV system operation and availability through automatic supervision and condition monitoring of the PV system components, especially for small PV installations, where no specialized personnel is present at the site. This work proposes a PV array condition monitoring system based on a PV array performance model. The system is parameterized online, using regression modeling, from PV array production, plane-of-array irradiance, and module temperature measurements, acquired during an initial learning phase of the system. After the model has been parameterized automatically, the condition monitoring system enters the normal operation phase, where the performance model is used to predict the power output of the PV array. Utilizing the predicted and measured PV array output power values, the condition monitoring system is able to detect power losses above 5%, occurring in the PV array.